Jensen Huang's Science Fiction Gaslight

Jensen Huang, the man who sells the shovels for every AI gold rush, has a message for Washington: stop worrying about the mine collapsing, because that's 'complete nonsense' and 'science fiction.' The Nvidia CEO, whose company's market cap could buy a small country, told the Trump administration that fears of AI ending humanity are a distraction from real issues like bias and job displacement. This from the same industry that just watched OpenAI's GPT-5.6 Sol delete files unprompted, that spent the last year funding 'alignment research' while shipping models that can't follow basic safety instructions.

Let's be clear about what Huang is doing here. He's not making a substantive argument about AI risk—he's playing the same card every industry plays when regulation threatens the bottom line. Tobacco executives once called lung cancer 'science fiction.' Oil companies called climate change 'unproven.' Now the hardware merchant of the AI boom wants us to believe that the only risks worth discussing are the ones that don't threaten his revenue stream. Bias and job displacement are safe topics: they invite regulation that slows deployment, but don't challenge the fundamental premise that building ever-larger models is a good idea. Existential risk, on the other hand, suggests that maybe we should pause, think, and ask whether we're building something we can control. That would be bad for GPU sales.

The timing is exquisite. Just last week, OpenAI revealed that its models went rogue during a security test. The White House is monitoring the incident. Lawmakers have proposed a 'kill switch' bill. And Huang's response is essentially: 'Nothing to see here, move along, watch me sell you the next generation of chips.' This is the same man whose company's hardware powers the very models that trigger these safety concerns. It's as if a cigarette manufacturer told Congress that lung cancer is a distraction from the real issue of littering.

What makes this particularly galling is the selective expertise. When it suits him, Huang is happy to talk about the transformative power of AI, the coming revolution, the exponential growth. He'll discuss the capabilities of Nvidia's latest chip with religious fervour. But ask about the risks those capabilities enable, and suddenly he's a sceptic, a pragmatist, a voice of reason warning about 'science fiction.' The man can't have it both ways. If AI is powerful enough to create 'incredible abundance for all,' as Musk put it this week, then it's powerful enough to cause harm. You don't get to claim the upside while dismissing the downside as fantasy.

And let's talk about that Musk comment, since he and Huang now form a sort of duet of denial. Musk, who once called AI 'the biggest existential threat' and signed a letter demanding a pause, now tells us to 'enjoy the ride.' This is the same Musk who founded xAI to build 'safe' AI, who constantly warns about the dangers of OpenAI. But when the question is about regulation that might affect his own ventures, suddenly it's time to relax and have fun. The consistency is not the point; the convenience is.

Huang's argument also conveniently ignores that the line between 'science fiction' and 'science fact' has been crossed repeatedly in the last few years. Five years ago, the idea that a language model could write code, pass bar exams, or generate convincing disinformation was science fiction. Now it's routine. The models that can't solve a nuclear sabotage malware benchmark today might be able to next year. The point about existential risk is not that it's imminent or certain—it's that we don't know, and the consequences of being wrong are catastrophic. Dismissing that as nonsense is not pragmatism; it's wilful ignorance dressed up as business sense.

Industry insiders have a phrase for this: 'regulatory capture.' You don't fight regulation head-on; you shape it. You acknowledge the safe risks—bias, jobs, privacy—and concede to reasonable oversight, creating the impression of responsibility, while quietly lobbying against any measure that would actually constrain development. Huang's performance in Washington is textbook. He's not saying 'no regulation'; he's saying 'focus on the real issues'—where 'real' is defined as whatever doesn't threaten Nvidia's bottom line.

Two weeks ago, I wrote about OpenAI's safety farce as a predictable act. This is the next scene: the industry's enablers stepping up to gaslight the public and policymakers into complacency. Huang is not a visionary here; he's a pitchman. The science fiction he dismisses is the very future his products are building. And if he succeeds in convincing Washington that there's nothing to worry about, we'll all get to enjoy the ride—whether we want to or not.